1: %\documentstyle[12pt,aasms4,psfig]{article}
2: \documentclass[12pt,preprint]{aastex}
3: \begin{document}
4: \title{GALEX and Optical Light Curves of EF Eridanus During a Low State: the Puzzling Source of UV Light}
5:
6: \author{Paula Szkody\altaffilmark{1},
7: Thomas E. Harrison\altaffilmark{2},
8: Richard M. Plotkin\altaffilmark{1},
9: Steve B. Howell\altaffilmark{3},
10: Mark Seibert\altaffilmark{4},
11: Luciana Bianchi\altaffilmark{5}}
12: \altaffiltext{1}{Department of Astronomy, University of Washington,
13: Box 351580,
14: Seattle, WA 98195, szkody@astro.washington.edu,plotkin@astro.washington.edu}
15: \altaffiltext{2}{Department of Astronomy, New Mexico State University, Box 30001, MSC 4500, Las Cruces, NM 88003, tharriso@nmsu.edu}
16: \altaffiltext{3}{WIYN Observatory and National Optical Astronomy Observatories,
17: 950 N. Cherry Avenue, Tucson, AZ 85726, howell@noao.edu}
18: \altaffiltext{4}{California Institute of Technology, MC 405-47, Pasadena, CA 91125, mseibert@srl.caltech.edu}
19: \altaffiltext{5}{Center for Astrophysical Sciences, The John Hopkins University,
20: 3400 N. Charles St., Baltimore, MD 21218, bianchi@skysrv.pha.jhu.edu}
21:
22: \begin{abstract}
23: Low state optical photometry of EF Eri during an extended low accretion state
24: combined with GALEX near and far UV
25: time-resolved photometry reveals
26: a source of UV flux that is much larger than the
27: underlying 9500K white dwarf, and that is highly modulated on the orbital period.
28: The near UV and
29: optical light curves can be modeled with a 20,000K spot but no spot model can
30: explain both
31: the large amplitude FUV variations and the SED. The limitations of limb
32: darkening, cyclotron and magnetic white dwarf models in explaining the
33: observations are
34: discussed.
35: \end{abstract}
36:
37: \keywords{binaries: close --- binaries: spectroscopic ---
38: cataclysmic variables --- ultraviolet: stars}
39:
40: \section{Introduction}
41:
42: Accreting close binaries containing white dwarfs with high (10-200 MG)
43: magnetic fields are termed AM Herculis variables or simply polars (review
44: in Wickramasinghe \& Ferrario 2000). They tend to have lower mass transfer
45: rates than disk accreting systems of similar orbital period and spend perhaps
46: the majority of their time (Ramsay et al. 2004)
47: in states where the mass transfer can drop by orders of magnitude
48: or stop
49: completely (termed low states). During these times, most of the effects
50: associated with active accretion (X-rays, strong optical cyclotron continuum,
51: strong He and Balmer emission lines) disappear. This provides the opportunity to
52: directly view and study the underlying stars (white dwarf and late type secondary).
53: The polar cataclysmic variable EF Eri received notoriety
54: as it remained in one of these inactive states for 9 years, starting in 1997
55: (Wheatley \& Ramsay 1998; WR98) and only becoming active again at the beginning of
56: 2006. In the high state, EF Eri is 14th mag and shows a highly variable
57: light curve over the 81 min orbital period (Bailey et al. 1982). In the
58: infrared, the J light curve revealed a narrow dip which Bailey et al. used
59: to define phase 0.0. At this phase, the $UBV$ light curves showed a broad
60: minimum,
61: the X-ray had a sharp minimum and circular polarization was
62: a maximum. These features were interpreted as an eclipse of the accretion
63: column by the accretion stream at phase 0. Cyclotron humps led to estimates
64: of 16.5 and 21 MG for two cyclotron regions.
65:
66: When EF Eri entered its prolonged low state in 1997, the V mag dropped to
67: 18, broad Balmer absorption lines from the white dwarf were apparent, and
68: Zeeman splitting indicated a magnetic field of 14 MG (WR98). While
69: Euchner et al. (2003) fit multipole models (5 components with fields up to
70: 100MG) to spectra obtained in 2000, the fit was not definitive. WR98 concluded
71: the
72: similarity of the low state field
73: to that measured in the small accretion regions during the high state
74: indicated either a uniform field strength over the white dwarf, or continued
75: accretion heated spots.
76: Beuermann et al.
77: (2000) modeled the optical spectrum with a
78: 9500K white dwarf with a 15,000K hot spot covering 6\% of the surface.
79: $BVRI$ photometry obtained by Harrison et al. (2003; H03) in 2001, showed
80: sinusoidal light curves with a peak near phase 0.9 and a minimum near
81: phase 0.4. They modeled the light curves with a 12,000K spot covering 6\% of
82: the area of a 9500K white dwarf,
83: with a best fit obtained for an inclination of 35$^\circ$ and an angle of 35$^\circ$ between the spin
84: and magnetic pole. The $H$ and $K$ light curves were anti-phased from the
85: optical, with a peak near
86: phase 0.5 and a minimum near phase 0. IR spectra obtained in 2002 (Harrison
87: et al. 2004; H04) showed different cyclotron harmonics were present at phase 0
88: than at phase 0.5, indicating that cyclotron emission was still present from both
89: accretion poles.
90:
91: Given that the optical spectrum of EF Eri at the low state shows no
92: obvious sign of an accretion stream, it
93: is intriguing that a hot spot model still fits the optical
94: light curves and cyclotron is still present in the IR. During a study of polars at low states, Araujo-Betancar (2005) found that 30,000-70,000K hot spots
95: contributing 20-30\% of the white dwarf flux were needed to model the FUV fluxes. To
96: further explore the heating effects at extremely low accretion rates,
97: we obtained UV light curves with GALEX. Our results show the value of GALEX
98: time-resolved photometry in providing information confirming
99: that some area of the white dwarf contributes a large amount of
100: UV flux even after 7 years of
101: extremely low mass transfer.
102:
103: \section{Observations}
104:
105: GALEX observations were first obtained on 2004 November 8 during 10
106: contiguous satellite orbits from 04:32:38 to 19:47:50 UT.
107: The {\it GALEX} satellite (Martin et al. 2005) uses a dichroic to split the
108: UV light into a FUV detector (1350-1750\AA) and a NUV detector (1750-2800\AA).
109: The field of view is 1.25$^\circ$ with 5 arcsec resolution in the FUV and 6.5 arcsec
110: in the NUV. During this observation, elevated solar proton levels resulted
111: in the FUV detector being switched off, so only data in the NUV were obtained.
112: The observation was rescheduled and the full 10 orbits with both detectors
113: were accomplished on 2004 December 7 from 03:46:02 to 18:59:40 UT.
114:
115: While the standard GALEX data pipeline converts time-tagged photons into
116: a final calibrated image for each observation,
117: the production of light curves required the generation of calibrated images in
118: 120s intervals\footnote{Produced by the GALEX Software Operations and Data
119: Analysis team.}, which were then
120: phased according to the
121: ephemeris of Bailey et al. (1982).
122: The IRAF\footnote{{IRAF (Image
123: Reduction and Analysis
124: Facility) is distributed by the National Optical Astronomy Observatories, which
125: are operated by AURA,
126: Inc., under cooperative agreement with the National Science Foundation.}}
127: routine {\it qphot} was then used to obtain a magnitude for the photons inside
128: a 9 pixel aperture
129: at the source position. The background was
130: measured in an annulus of width 3 pixels around the aperture. The
131: conversions from cps to magnitude to flux were accomplished using the
132: values from the GALEX online documentation\footnote{
133: http://galexgi.gsfc.nasa.gov/tools/index.html}
134: where FUV m$_{0}$=18.82=1.40$\times$10$^{-15}$ ergs cm$^{2}$ s${-1}$ \AA$^{-1}$
135: and NUV m$_{0}$=20.08=2.06$\times$10$^{-16}$ ergs cm$^{2}$ s${-1}$ \AA$^{-1}$.
136: To insure that the optical light curve had not changed from 2001 December
137: (H03), further photometry was obtained on 2005 September 16.
138: As in 2001, the New Mexico State University 1.0 m telescope was used with
139: a CCD and Johnson-Cousins V filter. A differential light curve was obtained
140: with respect to nearby stars on the same frames and Landolt standards were
141: used to calibrate the reference stars.
142: The combined GALEX and V light curves are shown in Figure 1.
143:
144: \section{Light Curve Modeling with a Hot Spot}
145:
146: As shown in Figure 1, the FUV, NUV, and $V$ light curves show sinusoidal-like
147: variations that appear to have similar phasing to each other and to the
148: $BVRI$ light curves of H03. The symmetrical nature and width of the variation
149: implies large and symmetrical regions of emission.
150: As with the optical light curves of H03,
151: they are all anti-phased to the low state
152: $H-$ and $K$-band light curves shown in H04 (the $J$-band light curve is more
153: complex). The amplitude of the variations in the optical are 0.2 mag peak-to-peak, while
154: the UV and IR variations have considerably larger amplitudes (up to 0.8 mag in
155: FUV and $K$). The time-resolved IR spectra of H04
156: confirmed that the $H-$ and $K$-band variations were due to cyclotron
157: emission.
158:
159: H03 showed the optical variations can be easily modeled using a simple hot spot
160: on a
161: 9500 K white dwarf. They
162: used the light curve modeling program WD98 (Wilson 1998) to confirm that
163: such a model could explain both their $BVRI$ light curves and the optical
164: spectral energy distribution (SED). Presumably, this hot spot would be located
165: near one of the magnetic poles, and represent the heating induced by continued
166: accretion whose presence is clearly demonstrated by the strong cyclotron
167: emission observed during this low state.
168:
169: Given their identical phasing, we attempted to use the same hot spot model
170: to explain the GALEX light curves of EF Eri. However, it soon became apparent
171: that
172: a 14,000 K hot spot on a 9,500 K white dwarf does not provide sufficient
173: ultraviolet flux at short wavelengths to simultaneously explain the observed
174: NUV and FUV fluxes. To achieve a reasonable fit to the observed FUV/optical
175: SED requires that the hot spot have a temperature of T$_{\rm eff} \geq$ 20,000
176: K. Figure 2 shows the spectral energy distributions (SEDs) for the GALEX and
177: optical data for
178: minima (phase 0.4; filled dots) and maxima (phase 0.9; stars) light (the values used are listed in Table 1). The
179: fit to the minimum SED (green line) is for a 9800K white dwarf (red dotted line)
180: that contributes 78\% of the V flux, with the rest from a 22,000K hot spot
181: (blue dotted line). It is obvious that to account for the light curve
182: maxima will require a larger contribution of a component near 20,000K.
183:
184: With this insight, we constructed a large number of models using
185: WD2005\footnote{WD2005 is an updated version of WD98, and can be obtained at
186: this website maintained by J. Kallrath:
187: http://josef-kallrath.orlando.co.nz/HOMEPAGE/wd2002.htm} with differing spot
188: sizes, orbital inclinations and spot latitudes. For our trial runs, none of our
189: models included limb darkening.
190: Our
191: initial models had a single spot temperature. Given the large number of
192: adjustable parameters, it was easy to generate a model that could fit the
193: $V$ and NUV light curves.
194: Figure 3 shows a model
195: for an inclination of 45$^\circ$ and a 20,000K spot with
196: a radius of 60$^\circ$
197: located 10$^\circ$ from the spin pole. However, no single temperature spot model
198: could explain the large amplitude of the FUV light curve--even unrealistic
199: models where one entire hemisphere was at 20,000 K, and those types of models
200: could not simultaneously explain the UV/optical SED.
201:
202: The difficulty with the single temperature spot models is that they must
203: supply both an enormous FUV flux $and$ produce $~\sim$ 1 magnitude variations.
204: For example, a 9500 K white dwarf normalized to the $V$-band flux,
205: only supplies about 4\% of the observed FUV flux. Thus, this 20,000K spot must
206: provide the remaining flux in the FUV at all orbital phases, and somehow be
207: highly variable. We thus explored more complicated models, ones in which the
208: temperature of the ``core'' of this spot was very high (T$_{\rm eff}$ = 100,000
209: K), which had cooler, off-centered annuli surrounding it that smoothly dropped
210: off in temperature with radius until they reached the observed photospheric
211: temperature of the white dwarf. Even these models were unable to produce both
212: the large amplitude FUV variations and the observed SED.
213:
214: One set of spot models that could reproduce the observed
215: light curve variations was models with extreme FUV limb darkening. Using a 20,000 K
216: spot, models with extreme limb darkening in the FUV (i.e. a coefficient of 2.0, a
217: factor of 2 over normal values for hot stars),
218: while the NUV
219: and V-band limb darkening coeffients had normal values of 0.5 and 0.3
220: (Al-Naimiy 1987, Wade \& Rucinski 1985) could produce large amplitude variations.
221: Figure 4 shows this type of model fit to the GALEX and V data.
222: While we cannot fully rule out such
223: models, such an abrupt change in limb
224: darkening laws between the FUV and NUV bandpasses is unlikely, especially given
225: the
226: expectation of few large spectral line features in the GALEX bandpasses. It is
227: interesting to note that Diaz et al. (1996) found quite large limb darkening
228: effects for accretion disks in the UV, depending on the line-of-sight
229: inclination angle of the disk. A tilted, flat surface like an accretion disk
230: has larger limb darkening than a spherically symmetric star since the
231: emergent flux from a spherical object is averaged over all angles. Thus, it does
232: not seem impossible to envision a larger FUV limb darkening for a relatively
233: small, isolated spot. If one were to add some vertical extent to this spot, its
234: limb darkening could be quite large. But, no matter what limb darkening is used, such a
235: spot/region needs to have a spectrum that preferentially produces FUV emission.
236:
237: \section{Discussion}
238:
239: While it is now evident that a source of far UV photons remains on the white dwarf during
240: extremely low states of accretion, the nature of this component is not clear.
241: G\"ansicke et al. (2006) were able to successfully fit far UV (FUSE) and near UV (STIS)
242: light curves of AM Her during a low state using a hot spot of
243: 34,000-40,000K on a 20,000K white
244: dwarf. In comparison to EF Eri, their bandpass closest to our
245: GALEX FUV one shows amplitudes that are about half those for EF Eri and the
246: hotter
247: white dwarf and spot can account for the peak amplitudes and fluxes near
248: 1100\AA. However, they did not see any change in temperature of the hot
249: spot in AM Her over the course of several months, and puzzled over the possible
250: sources of heating such as low ongoing accretion during the low state
251: or deep heating of the
252: pole cap area during times of high accretion.
253:
254: For EF Eri, the results of H04 and the shapes of the light curves offer some
255: clues to other possibilities. The IR cyclotron features at the low state indicate two poles with different
256: magnetic fields (if other poles are present, they are weak compared to these).
257: The opposite behavior of J vs H and K indicates J has a
258: cyclotron feature that is likely from a different accreting pole during the
259: high state (evident at photometric phase 0), whereas the low state shows cyclotron from the pole evident near phase 0.5. It is interesting that both the GALEX and V
260: data (Figure 1) show a disturbance in the light curves near phase 0.5, that
261: could indicate cyclotron is somehow playing a role in the UV. The
262: J band light curve at the low state also shows a slight peak at this phase.
263: For field strengths such as present in EF Eri, the cyclotron harmonics are
264: usually optically thick in the IR and optically thin in the optical and UV.
265: The thin harmonics would be best seen perpendicular to the magnetic field,
266: thus explaining the offset in phase from the opt/UV to the IR, if cyclotron
267: was responsible.
268:
269: The optical light curves of the Low Accretion Rate Polars (LARPS) show very similar
270: sinusoidal variability (Szkody et al. 2003) that can be explained by the
271: presence of cyclotron harmonics in the bandpass of observation. Thus, if
272: there were a cyclotron component in the UV, it could account for the
273: extra UV light. However, for the field strengths listed, it would be
274: difficult to get a component only in the UV that is not present in the
275: optical.
276: Schwope, Schreiber \& Szkody (2006) have proposed cyclotron components in
277: the UV to explain the discrepancy between the SED and model WD in the polar
278: RXJ1554.2+2721. That system has a field strength of 110 MG and
279: shows noticeable cyclotron humps
280: in the optical. While Euchner et al. (2003) proposed multipoles up to these
281: field strengths for EF Eri, their spectra showed no visible cyclotron humps.
282: However, a recent Keck spectrum taken in 2006 January, as EF Eri was emerging
283: from its long low state (Howell et al. 2006), shows possible harmonics near
284: 9300\AA\ and 4700\AA\
285: which are consistent with a 115 MG field.
286:
287: Schmidt et al. (2005) point out the discrepancy in LARPS between the
288: higher observed blue flux compared to white dwarf models that fit the longer
289: wavelengths.
290: This same discrpancy was also shown in modeling the UV flux of the highest field
291: polar AR UMa (G\"ansicke et al. 2001) and ascribed to the lack of known
292: correct models for high magnetic field white dwarfs. Thus, it is possible
293: that the discrepancy could go away if realistic models show a steeper
294: flux distribution for the white dwarf areas near the magnetic poles.
295: Modeling also suffers from correct treatment of the irradiated atmospheres of
296: white dwarfs. K\"onig, Beuermann \& G\"ansicke
297: (2006) have recently explored the atmosphere around a polar cap
298: irradiated by bremsstrahlung and cyclotron flux. For low magnetic field systems,they find the irradiated areas will be large and the reprocessed
299: energy will be in the FUV. While their models are for higher accretion rates
300: than in systems like EF Eri and LARPS, and they point out
301: that a much larger effort will be required to model the lowest accretors
302: correctly, it is a step in the right direction.
303:
304: Whether the extra
305: UV flux is due to hot spots, cylotron harmonics or incorrect UV models
306: for high field white dwarfs
307: remains to be determined.
308:
309: \section{Conclusions}
310: Our results show that GALEX time-resolved photometry is a very useful
311: means of studying the hot components in close binaries.
312: Our near and far UV light curves obtained during the extended low
313: accretion state of EF Eri reveal the presence of a symmetrical, large and
314: highly variable source of
315: UV light. If this is a
316: hot spot on the 9500K white dwarf, this spot
317: must have a temperature $>$20,000K to fit the SED. However, the amplitude
318: of the FUV variation is far too large to be fit with a spot of this type,
319: unless large limb darkening is invoked just for the FUV. Since the UV light
320: curves are in phase with the optical, and opposite in phase from IR
321: variations that are known to originate from cyclotron harmonics, it is natural
322: to invoke changing optical depth of harmonics from optically thick in the
323: IR to optically thin in the optical/UV. If there was a harmonic present in
324: the far UV band, it could explain the excess FUV flux and amplitude, but this
325: would require extremely high fields and harmonics should be visible in the
326: optical as well. It is intriguing that small deviations
327: in the UV and optical light curves match the phases of peak IR flux and that the
328: excess UV luminosity over that of the 9500K white dwarf is comparable to the
329: cyclotron luminosity in the IR. However, the lack of good models for
330: cyclotron effects on the spectrum of a white dwarf
331: prevents us from reaching any definite conclusion.
332: Until we can correctly model the
333: observed excess UV flux in the cases of high magnetic field white dwarfs,
334: we cannot ascertain its cause as a hot spot or cyclotron emission or
335: atmospheric effects due to high fields and low accretion.
336:
337: \acknowledgments
338:
339: Support for this research was provided by NASA GALEX grant NNG05GG46G.
340:
341: \begin{references}
342:
343: \reference{} Al-Naimiy, H. M. 1978, \apss, 53, 181
344:
345: \reference{} Araujo-Betancor, S., G\"ansicke, B. T., Long, K. S., Beuermann, K., de Martino, D., Sion, E. M. \& Szkody, P. 2005, \apj, 622, 589
346:
347: \reference{} Bailey, J., Hough, J. H., Axon, D. J., Gatley, I., Lee, T. J., Szkody, P., Stokes, G. \& Berriman, G. 1982, \mnras, 199, 801
348:
349: \reference{} Beuermann, K., Wheatley, P., Ramsay, G., Euchner, F. \& G\"ansicke,
350: B. T. 2000, \aap, 354, L49
351:
352: \reference{} Diaz, M. P., Wade, R. A., \& Hubeny, I. 1996, ApJ, 459, 236
353:
354: \reference{} Euchner, F., Beuermann, K., Reinsch, K. \& Jordan, S. 2003, in
355: White Dwarfs, eds. D. de Martino, R. Silvotti, J. E. Solheim \& R. Kalytis, Kluwer, 195
356:
357: \reference{} G\"ansicke, B. T., Schmidt, G. D., Jordan, S., \& Szkody, P. 2001, \apj, 555, 380
358:
359: \reference{} G\"ansicke, B. T., Long, K. S., Barstow, M. A. \& Hubeny, I. 2006,
360: \apj, in press
361:
362: \reference{} Harrison, T. E., Howell, S. B., Huber, M. E., Osborne, H. L., Holtzman, J. A., Cash, J. L. \& Gelino, D. M. 2003, \aj, 125, 2609 (H03)
363:
364: \reference{} Harrison, T. E., Howell, S. B., Szkody, P., Homeier, D., Johnson, J. J. \& Osborne, H. L. 2004
365: , ApJ, 614, 947 (H04)
366:
367: \reference{} Howell, S. B., Walter, F., Harrison, T. E., Huber, M. E., Becker, R. H.
368: \& White, R. L. 2006, ApJ, submitted.
369:
370: \reference{} K\"onig, M., Beuermann, K., \& G\"ansicke, B. T. 2006, \aap, in press
371:
372: \reference{} Martin, D. C. et al. 2005, \apj, 619, L1
373:
374: \reference{} Ramsay, G., Cropper, M., Wu, K., Mason, K. O., Cordova, F. A. \&
375: Priedhorsky, W. 2004, \mnras, 350, 1373
376:
377: \reference{} Schmidt, G. D. et al. 2005, \apj, 630, 1037
378:
379: \reference{} Schwope, A. D., Schreiber, M. R. \& Szkody, P. 2006, \aap, in press
380:
381: \reference{} Szkody, P. et al. 2003, \apj, 583, 902
382:
383: \reference{} Wade, R. A. \& Rucinski, S. M. 1985, \aaps, 60, 471
384:
385: \reference{} Wheatley, P. J. \& Ramsay, G. 1998, in ASP Conf. Ser. 137, Wild
386: Stars in the Old West, ed. S. Howell, E. Kuulkers, \& C. Woodward (San FranciscoLASP), 446 (WR98)
387:
388: \reference{} Wickramasinghe, D. T. \& Ferrario, L. 2000, \pasp, 112, 873
389:
390: \reference{} Wilson, R. E. 1998, Reference Manual to the Wilson-Devinney Program,
391: Computing Binary Star Observables, Version 1998 (Gainsville; Univ. Florida)
392:
393: \end{references}
394:
395: \clearpage
396:
397: \clearpage
398: \begin{deluxetable}{ccc}
399: \tablewidth{0pt}
400: \tablecaption{Mean Mags at Bright \& Faint Phases}
401: \tablehead{
402: \colhead{Wavelength(\AA)} & \colhead{Phase 0.9} & \colhead{Phase 0.4}}
403: \startdata
404: 1528 & 18.58 & 19.30 \\
405: 2271 & 18.37 & 18.70 \\
406: 4360 & 18.04 & 18.37 \\
407: 5450 & 18.04 & 18.21 \\
408: 6380 & 17.90 & 18.08 \\
409: 7970 & 17.76 & 17.94 \\
410: \enddata
411: \end{deluxetable}
412:
413: \clearpage
414: \begin{figure} [h]
415: \figurenum {1}
416: \plotone{f1.eps}
417: \caption{GALEX FUV (top), NUV (mid) and V (bottom) light curves as a function of
418: Bailey phase. Error bars are shown for one cycle in the GALEX plots and
419: the V light curve shows 2 sequential orbits which are repeated. Error bars
420: in the V data are $\pm$0.07mag.}
421: \end{figure}
422:
423: \begin{figure}
424: \figurenum {2}
425: \plotone{f2.eps}
426: \caption{SED for GALEX through optical wavelengths. Filled circles are
427: minima light (phase 0.4), stars are maxima light (phase 0.9), solid green line is
428: the sum of a fit of a 9800K blackbody contributing 78\% of the V flux (right red
429: dashed curve) and a 22000K blackbody (left blue dashed curve). Note that the
430: fit of the maxima light would require a larger contribution of the 22000K BB.}
431: \end{figure}
432:
433: \begin{figure}
434: \figurenum {3}
435: \plotone{f3.eps}
436: \caption{Best fit spot model to the GALEX NUV and optical V light curves
437: with a 9800K white dwarf and a 20,000K hot spot located 10$^{\circ}$ from
438: the spin axis with an inclination of 45$^{\circ}$. Solid, dotted, short
439: dash, long dash, short dash-dot, and long dash-dot curves correspond to
440: spot radius of
441: 5,10,20,30,45 and 60$^{\circ}$.}
442: \end{figure}
443:
444: \begin{figure}
445: \figurenum {4}
446: \plotone{f4.eps}
447: \caption{Best fits to GALEX and V light curves with a 20,000K spot model and
448: a large FUV limb darkening (limb darkening coefficients of 2.0, 0.5 and 0.3 for
449: FUV, NUV and V).}
450: \end{figure}
451:
452: \end{document}
453: